When used in an optical computer or communications switch, such devices could implement a variable-length optical delay line. This long-elusive goal of EEs would be enabled by the smooth variation of the speed of light in the fiber, rather than by changing the length of the delay line, which is the ordinary solution.
"We found a surprisingly simple rule of thumb: You get one nanosecond of delay for each decibel of gain you introduce with Brillouin scattering," said Thevenaz.
Gain-which in this case is narrowband amplification at a sharp spectral transition-controls the speed of light when the Brillouin scattering technique is used. Brillouin scattering is stimulated by the interaction of two counterpropagating waves-a pump wave and a probe wave. If the phase of the pump wave matches closely the probe wave, then any slight differences create an acoustic wave that scatters photons from the pump to the probe, thereby moving energy within the spectrum to achieve amplification over a narrow bandwidth. However, while the real parts of the amplification involve gain, the phase changes in the imaginary parts of the amplification create a faster-than-light mechanism.
Usually, the speed of light in a vacuum is fixed at about 300,000 kilometers per second (or about 186,000 miles per second), but by varying the gain from 4.26 down to zero and then beyond to -0.7, the researchers were able to vary the speed of light from 205,000 meters/second to zero and then to -428,000 meters/s (called a Brillouin loss). The negative index during Brillouin loss means that the pulse peak exits the fiber before entering it (in the narrow band). Consequently, the variable delay could be tuned from -14.4 nanoseconds to 18.6 nanoseconds in only 2 meters (or about 6.6 feet) of fiber. In terms of length, that made the effective fiber length continuously variable from -3 meters to 3.8 meters.
"The kind of light speed we strongly modified was the 'group velocity' that in textbooks is associated with the speed of the signal. But that is wrong for the special case of narrowband absorption or amplification at a sharp spectral transition. There the speed of information no longer corresponds to the group velocity, but remains bounded by c, as stated by relativity, and no experiment, including ours, contradicts this prediction so far," said Thevenaz.